Mitigating Epistemic Uncertainty in Structural Identification: Case Study for a Long-Span Steel Arch Bridge

Abstract

Characterization of constructed civil-engineering systems through structural identification (St-Id) has gained increasing attention in recent years due to its potential to enable more effective infrastructure asset management and performance-based engineering. Although there have been recent advances that mitigate the challenges posed by aleatory (random) uncertainty, there are many remaining challenges associated with epistemic (bias) uncertainty that often have a more critical impact on the reliability of St-Id (especially with applications to constructed systems). The objective of this paper is to illustrate various sources of epistemic uncertainty and describe mitigation approaches by detailing the St-Id of a long-span steel arch bridge. This application includes a priori modeling, ambient vibration monitoring, data processing, feature extraction, and finite-element (FE) model correlation. Following a description of the St-Id, the impact of various modeling uncertainties on the calibrated FE model is evaluated by comparing different identification scenarios. Finally, a simple and rational approach to “truth testing” the identified model is developed and employed to determine its admissibility.